Medical Image Analysis & Machine Learning Applications
Here we investigate image analysis and machine learning techniques and their application for clinical medical tasks like segmentation and computer-aided detection.
In this project we are interested in automatically deriving an estimate of age of adolescents, mainly from MR images. This includes localization of age-related anatomical structures and regressing skeletal or chronological age from training data.
Here we investigate computer vision and 3D visualization techniques to post-process and illustrate forensically relevant evidence as derived from radiological imaging methods (CT, MRI), with the ultimate goal of presenting evidence in court in a self-explanatory manner.
Tubular Structures & Lung Image Analysis
For the analysis of vascular structures image processing methods based on tubularity assumptions have been investigated. More recently, vessel extraction in the lung as well as artery/vein separation and lung lobe segmentation became topics of interest.
We have contributed in a number of nonlinear registration approaches mostly for studying lung motion from CT during breathing. Investigated methods involved feature matching using SIFT and shape context in an intensity-based framework as well as TV based registration.
In a collaboration with the Siemens IT Solutions and Services Biometric Center, Graz we have developed algorithms for analyzing facial images for their suitability in ICAO conform documents (e.g. passports), regarding facial expression and facial pose.
In a collaboration with the Siemens IT Solutions and Services Biometric Center, Graz we have developed algorithms for analyzing fingerprint images and ridge orientation models with the aim of improving existing fingerprint biometry technology.